How to ask AI questions about your bills and documents

You have a stack of utility bills from the past year and you need to know how much you spent on electricity in the first quarter. Or there is a clause buried somewhere in your tenancy agreement about early termination, and you cannot remember whether it requires 30 or 60 days' notice. Maybe an old invoice from a contractor lists a warranty period, but you would have to dig through a folder of PDFs to find it.

These are not complicated questions. The information exists, sitting in documents you already have. The problem is that finding it means opening files one at a time, scanning through pages, and hoping you spot the right paragraph before your patience runs out.

Fabric's AI assistant lets you skip all of that. You type a question in ordinary language, the way you would ask a friend who happened to have perfect recall of every document you have ever saved, and get a direct answer drawn from your stored files.

How it works in practice

The experience is closer to a conversation than a search. Rather than constructing the right keyword combination and hoping it lands on the correct document, you simply ask what you want to know. "How much did I pay for electricity in Q1?" "What is the cancellation clause in my lease?" "When does the warranty expire on the boiler installation?"

Fabric's AI reads through your stored documents, locates the relevant sections, and returns an answer with references to the source material. You can follow up with clarifying questions, ask for comparisons across multiple documents, or request summaries of lengthy contracts. The assistant works across your entire collection, so a single question can pull information from a utility bill, an email confirmation, and a scanned letter all at once.

This works because of two things happening beneath the surface. First, Fabric processes your documents when you store them, extracting and indexing their contents so the AI can reference them quickly. Second, the memory engine builds an understanding of how your documents relate to one another. It recognises that a series of monthly bills comes from the same provider, that an email thread refers to the same contract sitting in your files, and that a handwritten note you scanned relates to a phone call about an insurance claim. This contextual awareness means the AI can give you answers that account for the full picture, not just isolated fragments from a single file.

Getting your documents into Fabric

The AI assistant can only answer questions about documents it has access to, so the first step is getting your files into Fabric's cloud storage. There are several ways to do this, depending on what kind of documents you are working with.

Digital files like PDFs, images, and downloaded statements can be uploaded directly or saved using quick capture, which lets you clip documents from your browser or other apps into Fabric without interrupting what you are doing. If your bills and receipts arrive by email, Fabric's email connections can bring them in from Gmail and other providers, so they appear in your collection alongside everything else.

Paper documents need one extra step. Fabric includes a handwriting scanner that captures paper bills, letters, and notes through your phone's camera, converting them into searchable, queryable documents. A utility bill that arrived in the post, a handwritten receipt from a tradesperson, or a letter from your council all become part of your searchable collection once scanned.

The AI assistant handles all of these formats, whether a document started as a typed PDF, a photograph of a receipt, an email attachment, or a scanned sheet of paper. You do not need to convert or prepare anything.

The kinds of questions you can ask

Because the assistant understands natural language, you are not limited to keyword lookups. You can ask broad or specific questions, and the assistant will interpret your intent and search accordingly.

Financial questions tend to be the most common starting point. "How much was my gas bill in March?" pulls the figure from the relevant statement. "What have I spent on insurance this year?" requires the assistant to gather amounts from multiple documents and add them together. "Has my broadband cost gone up since last year?" involves comparing figures across two billing periods. These are questions that would take you several minutes of file-hunting to answer manually, but the AI handles them in seconds.

Contract and agreement questions are where the assistant saves the most frustration. Lease agreements, insurance policies, service contracts, and subscription terms tend to be long, dense, and written in language designed to be precise rather than readable. Asking "What notice period do I need to give my landlord?" or "Does my home insurance cover accidental damage?" pulls out the specific clause without requiring you to read the entire document.

You can also ask questions that span multiple documents. "Which of my subscriptions increased their price this year?" requires the assistant to compare old and new billing statements across several services. "Do any of my contracts auto-renew?" scans through all your agreements to find renewal clauses. This kind of cross-document analysis is where the memory engine proves its value, since it already understands the relationships between your stored files.

Beyond simple lookups

The AI assistant does more than retrieve isolated facts. You can ask it to summarise a lengthy contract, explain a confusing paragraph in plain terms, or highlight the sections of a document that are most relevant to a specific concern. If you have received a confusing letter or bill, you can paste or scan it and ask the assistant to walk you through what it means.

This is particularly useful for legal and financial documents where the language is deliberately technical. Rather than searching for a definition of "consequential damages" or trying to parse a clause about indemnification, you can ask "What am I liable for if I cancel early?" and get an answer in conversational English.

The assistant also supports follow-up questions within the same conversation. If you ask about your electricity costs and the answer prompts a further question about whether your tariff changed, you can ask that as a follow-up without restating the context. The assistant remembers what you have been discussing and treats the conversation as a continuous thread.

Working with the AI through other tools

Fabric is not a closed system. Through MCP and agent access, you can reach your Fabric documents from tools like ChatGPT and Claude. If you already use one of these assistants for other tasks, you can connect it to Fabric and ask questions about your stored documents without switching apps.

Agent access is scoped and time-limited, meaning you control precisely what an external tool can see and for how long. You might grant a financial planning agent temporary access to your bills for a budgeting exercise, or let an AI assistant review your contracts for a specific project, without giving it permanent or unrestricted access to your entire collection.

Fabric's own agents can also process documents on your behalf. You could set up an agent to monitor incoming bills, flag unusual charges, extract key dates from contracts, or organise new documents as they arrive. This moves beyond asking questions one at a time and into ongoing, automated document intelligence.

Keeping your documents private

Asking an AI to read your financial documents and personal contracts raises an obvious question about privacy. Fabric addresses this directly. Your data is encrypted with AES-256 at rest, the platform holds CASA Tier 2 certification, and your documents are never used to train AI models. The AI processes your documents to answer your questions, but the contents remain private and secure within your account.

This matters because the documents people most want to query, bills, contracts, medical letters, legal correspondence, are also the documents they would least want exposed. Fabric is designed so that you get the convenience of AI-powered document access without compromising on data security.

Combining AI questions with search

The AI assistant and Fabric's semantic search complement one another. Search is best when you know a document exists and want to find it quickly, perhaps to download, share, or reference it directly. The AI assistant is better when you have a question and do not know which document contains the answer, or when the answer requires synthesis across several files.

In practice, you will use both. You might search for "council tax bill 2025" to pull up the specific document, then ask the AI "How does this compare to last year?" to get a quick analysis. Or you might start with an AI question, "Which provider gave me the cheapest quote?", and then use search to locate the full quote document once you have your answer.

Frequently asked questions

What types of documents can the AI assistant read?

The assistant works with PDFs, scanned documents, photographs of paper bills, email attachments, and most common file formats. If you can store it in Fabric, the AI can read it. This includes scanned documents captured through the handwriting scanner, so even paper-only correspondence becomes queryable once you photograph it.

How accurate are the answers?

The AI draws its answers directly from your stored documents and provides references to the source material, so you can verify any response. For numerical questions like bill amounts or dates, the assistant extracts the specific figures from your files. For interpretive questions about contract clauses, it provides both its interpretation and the original text so you can judge for yourself.

Can the AI compare information across multiple documents?

Yes. One of the most useful capabilities is cross-document analysis. You can ask questions like "How have my energy costs changed over the past six months?" or "Which of my insurance policies covers water damage?" and the assistant will search across all relevant documents to build its answer.

Do I need to organise my documents in a particular way?

No. The AI assistant and the memory engine work regardless of how your files are organised. You can have everything in a single folder or sorted into elaborate categories. The assistant searches based on content and context, not file location or naming conventions.

Can I ask questions about emails as well as documents?

If you have connected your email through Fabric's connections feature, then yes. Emails and their attachments become part of your searchable, queryable collection. You can save email bills and receipts automatically and then ask the AI questions about them alongside your other documents.

Is there a limit to how many documents the AI can search through?

Fabric's cloud storage is designed as an infinite cloud drive, so you can store and query as many documents as you need. The AI assistant searches across your entire collection when answering questions, regardless of size.

Can I use the AI assistant from ChatGPT or Claude?

Yes. Through MCP integration, you can access your Fabric documents from external AI tools. This means you can ask questions about your bills and contracts from whichever AI assistant you prefer, while your documents remain stored and secured in Fabric.

What happens to my data when I ask the AI a question?

Your documents are encrypted with AES-256, and Fabric holds CASA Tier 2 certification. When you ask a question, the AI processes the relevant document contents to generate an answer, but your data is never used to train AI models and remains private within your account.

Can agents automatically process my bills and documents?

Fabric's agents can be configured to handle documents as they arrive. An agent might extract due dates from bills, flag charges above a certain amount, or summarise new contracts as you save them. Agent access is scoped and time-limited, giving you control over what automated processes can see and do.

Which AI models does Fabric use?

Fabric's AI assistant is multi-model, drawing on providers including OpenAI, Anthropic, and Google. This means you benefit from the strengths of different models depending on the type of question you are asking, without having to choose or switch between them yourself.

Related pages

• How to use AI to explain a confusing letter or bill

• How to find an old bill or invoice fast

• How to search your scanned documents

• How to save email bills and receipts automatically